Jason Michael Hostetter MD, Presenter: Nothing to Disclose

James Jason Morrison MD, Abstract Co-Author: Nothing to Disclose

Jean Jeudy MD, Abstract Co-Author: Nothing to Disclose

Kenneth Chung-Yi Wang MD, PhD, Abstract Co-Author: Co-founder, DexNote, LLC

Eliot L. Siegel MD, Abstract Co-Author: Research Grant, General Electric Company Speakers Bureau, Siemens AG Board of Directors, Carestream Health, Inc Research Grant, XYBIX Systems, Inc Research Grant, Steelcase, Inc Research Grant, Anthro Corp Research Grant, RedRick Technologies Inc Research Grant, Evolved Technologies Corporation Research Grant, Barco nv Research Grant, Intel Corporation Research Grant, Dell Inc Research Grant, Herman Miller, Inc Research Grant, Virtual Radiology Research Grant, Anatomical Travelogue, Inc Medical Advisory Board, Fovia, Inc Medical Advisory Board, Toshiba Corporation Medical Advisory Board, McKesson Corporation Medical Advisory Board, Carestream Health, Inc Medical Advisory Board, Bayer AG Research, TeraRecon, Inc Medical Advisory Board, Bracco Group Researcher, Bracco Group Medical Advisory Board, Merge Healthcare Incorporated Medical Advisory Board, Microsoft Corporation Researcher, Microsoft Corporation

Pulmonary nodule cancer risk is correlated largely with nodule size in smokers and non-smokers as described by the Fleischner criteria, however risk varies widely within the size and density categories. Nodule risk predictions and surveillance strategies could be improved by incorporating multiple predictors into a tool to generate personalized matched NLST cohorts.

The Fleischner Society guidelines are the standard for follow-up of pulmonary nodules. Availability of clinical and image data from large clinical trials such as the National Lung Screening Trial (NLST) makes it possible to utilize “big data” methodology to further personalize these recommendations. We used the NLST dataset and a custom web interface to stratify pulmonary nodules based on nodule and patient attributes and can provide a data driven reference for personalized nodule cancer risk by creating matched cohorts.

Real-time analysis of the NLST dataset was performed through a custom web-based interface. Evaluating the nodule data using Fleischner size criteria alone predicted risk of malignancy of 2.8% for nodules <= 4 mm to 7.0% for nodules > 8 mm. Further analysis within each size category using discriminators such as nodule location and patient smoking history (pack years) provided additional risk stratification. For example, in patients with 4-6 mm pulmonary nodules, the risk of malignancy ranged from 1.9% to 8.6%. In patients with 6-8 mm nodules, risk ranged from 4.7% to 8.9%. The NLST dataset includes additional demographic and nodule descriptive data, and incorporating these will likely lead to even greater risk discrimination.

This work aims to evaluate and refine the Fleischner criteria using a large clinical dataset. The range of malignancy risk within Fleischner size categories suggests that size and nodule density are useful but incomplete predictors of nodule malignancy for a specific patient. Additional demographic and anatomic predictors of nodule malignancy have been identified in the literature. Leveraging matched NLST cohorts and incorporating additional cancer risk predictors could improve nodule follow-up recommendations.

Hostetter, J, Morrison, J, Jeudy, J, Wang, K, Siegel, E, Personalizing the Fleischner Criteria: Real Time Data Mining of a Large Clinical Trial Dataset.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045808.html